結合抽水與監測容量擴張之地下水量即時管理模式

Abstract

地下水系統的描述或數值模式，因各種簡化假設與水文地質參數而有不確定性存在。雖然愈準確之地下水系統描述，可以設計出更有效率之抽水策略，但仍需進一步評估監測資訊對於規劃抽水策略之助益是否合乎經濟成本。因此完備之地下水管理應同時考量抽水策略與監測策略之規劃設計，且應以容量擴張分階段設井型態，使得在即時操作過程中，經由監測資訊對系統狀態更新後，可再調整抽水與監測井網。雖然以往之地下水管理研究中亦有同時考量抽水量及監測井網，但並未考量抽水井網之最佳化，更無抽水井網與監測井網即時調整之相關研究。本研究目的為建立一地下水量即時管理模式，可隨系統之操作對抽水策略(抽水井網及抽水量)及監測策略(監測井網) 即時進行重新規劃調整，即時優選容量擴張型態之抽水策略及監測策略。 本研究整合遺傳演算法與退火演算法，優選容量擴張型態之井網及離散化時變抽水量，並以卡門濾波進行即時更新及水位誤差推估。利用遺傳演算法本身隱平行之特性，以分配染色體為計算量切割，在平行電腦大量計算資源下，克服所需大量計算。最後以一拘限含水層為簡例，驗證所發展之模式可應用於地下水量管理問題上。由數值結果顯示，經由考量抽水與監測相互影響機制所設置之監測井網，雖然佔總成本之比例並不高，但在地下水模擬模式經由監測資訊更新後，再重新調整井網，可有效降低總成本。

The uncertainty of simulation modeling is resulted from the simplified assumptions of reality and uncertainty of parameter estimation. With more information regarding aquifer characterization, the devised pumping strategy will be more cost-effective. The pumping and the monitoring strategies must be designed simultaneously, so that the trade-off between the cost of monitoring strategy and the value of information, i.e. pumping cost saving, could be appraised. For the flexibility of adjusting the well networks after updating the groundwater system, the well networks should be with capacity that expands with time. In this study, a real-time management model for jointly optimizing the pumping and monitoring stratgies of groundwater supply problem considering the system capacity expansion is presented. The proposed approcah hybridizes a Genetic algorithm (GA) and Simulation annealing (SA). The chromosomes of the GA represent a possible design alternative, a pumping and monitoring network with capacity that expands with time. The SA is then used to compute the optimal pumping policy associated with the chromosome. The information obtained through monitoring network is used to update the hydraulic head and which covariance matrix with Kalman filtering. The hybrid algorithm can be effectively parrallelized by the implicit parallel characteristic of GA. This study implements the parallel computation to overcome the increasing computational requirements due to the complexity of the problem as well as using GA and SA. A hypothetical confined aquifer is included to demonstrate the proposed apparoch and to illustrate the interaction between pumping and monitoring strategies. Results show that the ratio of monitoring cost to total cost is relatively small. Through integrating the pumping and monitoring designs and redesign the well networks with time, the monitoring strategy could significantly reduce the pumping cost.